Rehabilitation Psychology November 1999 Vol. 44, No. 4, 354-372 © 1999 by the Educational Publishing Foundation For personal use only--not for distribution. Depression and Subjective Well-Being Among 97 American Indians With Spinal Cord Injury A Descriptive Study J. Stuart Krause Shepherd Center Jennifer Coker Shepherd Center Susan Charlifue Craig Hospital Gale G. Whiteneck Craig Hospital ABSTRACT Objective: To measure both clinical and nonclinical aspects of depression and subjective well-being (SWB) among a sample of American Indians with spinal cord injury (SCI) and to identify factors associated with depression and SWB. Study Design: Interviews were conducted by telephone with all participants except those who did not have telephones (who returned materials by mail). Setting: A large rehabilitation hospital in the western—Rocky Mountain region of the United States. Participants: Ninety-seven American Indians with SCI completed a health-related interview that included measures of depression and SWB. All participants were adults with traumatic SCI of at least 1 year in duration, most of whom lived on reservations. Main Outcome Measures: Measures of biographic and injury status, depression, SWB, activities, and health-related behaviors. Results: The American Indian sample reported elevated levels of depression and diminished SWB in 5 of 8 areas relative to previous studies on non—American Indians with SCI. Hierarchical multiple regression accounted for 59% of the variance in depression scores and between 10% and 52% of the variance in SWB scales. Health status, activity patterns, social support, years of education, and frequency of alcohol use were among the primary predictors of the study outcomes. Conclusions: The results suggest that American Indians with SCI experience diminished SWB relative to other SCI populations and that interventions are needed to reduce alcohol misuse, to help build support networks, and to increase participation in social activities. Research on psychological outcomes among people with spinal cord injury (SCI) has suggested that it is important to look at several types of psychological outcomes. These include investigation of clinically relevant outcomes, particularly depression ( Fuhrer, Rintala, Hart, Clearman, & Young, 1993 ; Kemp & Krause, 1999 ; Kemp, Krause, & Adkins, 1999 ), as well as nonpathological outcomes, such as life satisfaction and post-SCI adjustment ( Carlson, 1979 ; DeVivo & Richards, 1992 ; Krause, 1992 , 1997 , 1998a ; Krause & Crewe, 1991 ). Both types of outcomes are relevant to the subjective well-being (SWB) of people with SCI. RACE—ETHNICITY AND SWB Despite a growing body of research on psychosocial outcomes after SCI, there continues to be little research measuring psychosocial outcomes among members of various racial—ethnic groups with SCI. The limited amount of existing research suggests that there are meaningful differences in psychological outcomes related to race—ethnicity. One study that has just been completed ( Kemp et al., 1999 ) has identified differences in depression as a function of race—ethnicity. One hundred seventy-one participants were administered the Older Adult Health and Mood Questionnaire (OAHMQ), a 22item true—false measure of depressive symptomatology designed to include a limited number of vegetative items that may overlap with the symptoms of SCI ( Kemp & Adams, 1995 ). No differences were identified between Caucasian and African American participants in terms of level of depression. However, Latino participants had a higher overall depression score (6.9) than both Caucasian (4.3) and African American (4.5) participants. A significantly greater percentage of Latino participants also had scores of 11 or higher, indicative of possible major depression (Latino: 26%; Caucasian: 15%; and African American: 0%). Studies focusing on more general SWB outcomes have suggested that African Americans with SCI report higher levels of problems and lower levels of satisfaction in particular areas of life. For example, in developing a 50-item factor-analytic measure of SWB, Krause ( 1998b , 1998c ) found that minority participants (most of whom were African American) reported lower SWB related to career opportunities, finances, and, to a lesser extent, home life. No racial—ethnic differences were noted on more global aspects of SWB such as life engagement, negative affect, and health; differences were observed with global measures that included items related to career, finances, and home life. An earlier study also showed that minority participants view themselves as having few opportunities to learn or to exercise job skills ( Krause & Anson, 1997a ). In summary, these studies have shown that minority participants report lower SWB outcomes in areas of life in which they may be disadvantaged by virtue of having fewer opportunities. Indeed, employment research has indicated that African Americans have lower employment rates ( DeVivo & Fine, 1982 ; DeVivo, Rutt, Stover, & Fine, 1987 ; James, DeVivo, & Richards, 1993 ; Krause & Anson, 1997b ; Krause, Sternberg, Maides, & Lottes, 1998 ). AMERICAN INDIANS AS A SPECIAL POPULATION There is little question that the culture and specific circumstances of American Indians deserve special attention in studies of long-term health and psychosocial outcomes. In fact, studies of general health have indicated that American Indians are a population at high risk for chronic disease ( Cheadle et al., 1994 ; Goldberg et al., 1991 ; Pearson, Cheadle, Wagner, Tonsberg, & Psaty, 1994 ; Peterson et al., 1994 ; Sugarman, Warren, Oge, & Helgerson, 1992 ; Welty et al., 1995 ). The life expectancy of American Indians has been found to be 3 years less than that of the general population, and their mortality rate is two times higher ( Goldberg et al., 1991 ). Miller (1978) found three possible explanations of why American Indians underuse health services: Very few American Indians perceive a need for services, many have a lack of knowledge about available services, and many experience cultural barriers to services. Little research has been done on aspects of psychosocial health and adjustment of this population. In one study on the prevalence of depression among American Indian adults from the general population who visited a primary care clinic, it was found that 20.7% scored positive for a depressive syndrome, whereas 8.9% met the criteria for a major depressive episode ( Wilson, Civic, & Glass, 1995 ). Depression in American Indian adults has also been linked to alcohol abuse, especially among women ( Lowe, Long, Wallace, & Welty, 1997 ). Even less research has been done on American Indians with SCI. In a recent study of health behaviors involving the same sample of American Indians who took part in this study, American Indians with SCI were less likely than the general population to rate their health as very good or excellent ( Krause, Coker, Charlifue, & Whiteneck, in press )(1999z). Also, American Indians with SCI reported significantly more days of poor physical health than the general population. American Indians with SCI were less likely to smoke and to have consumed alcohol in the previous month; among those who did consume alcohol, however, the average number of drinks per occasion was significantly higher. This is important because Morgan and O'Connell (1987) found that American Indians are more likely to have a disability related to alcohol or drug use, and, as mentioned earlier, depression has been linked to alcohol use ( Lowe et al., 1997 ). SUMMARY AND PURPOSE Taken together, existing research suggests that race—ethnicity may be related to differential SWB after SCI. The purpose of the current exploratory study was to identify descriptive data on both clinical and nonclinical psychological outcomes (i.e., depression and SWB) among American Indians with SCI and to identify factors related to these outcomes. In addition, the current findings were compared with scores on the same instruments identified with non—American Indian participants with SCI. Multiple regression was used to identify factors predictive of both clinical and nonclinical outcomes (i.e., depression and SWB). Although the current study was exploratory in nature and therefore only began to address issues in regard to American Indians with SCI, it was designed to lay the foundation for future studies that may examine cultural factors in relation to SWB outcomes as well as those that explore theory-based constructs in relation to post-SCI outcomes. We hope that generating data on the relationships of activity patterns, health behaviors, and social support with SWB after SCI will direct researchers toward the cultural and theoretical factors of greatest importance (e.g., alcohol misuse or availability of particular types of support). Study hypotheses were as follows. American Indians with SCI will report higher levels of depression than those previously reported by Caucasian and African American participants but levels similar to those of Latinos with SCI. American Indians with SCI will report lower levels of SWB related to specific areas of life (i.e., career opportunities and finances) in comparison with those previously reported in the literature, but no differences will be observed on general measures of life engagement, health, family life, or interpersonal relations. Social support will be the best predictor of family and interpersonal SWB. Activity patterns will be the best predictors of depression, life engagement, and negative affect. Income and years of education will be most predictive of SWB related to finances and career opportunities. METHOD Participants All participants were selected from files of a large rehabilitation hospital located in the western—Rocky Mountain area of the United States. There were five selection criteria for the study: (a) traumatic SCI, (b) some residual impairment from SCI (i.e., not a complete recovery), (c) occurrence of SCI no less than 1 year earlier, (d) age of at least 18 years at the time of the study, and (e) American Indian heritage. Of the 100 potential participants, 97 provided usable responses (97% response rate). Most of the participants were Navajo and Pueblo-Zuni, 12 were Sioux, and 4 were of mixed ancestry. All were from the western United States, most living on reservations in Colorado. Seventy-two participants completed the interview by telephone, whereas 25 did not have telephones and completed and returned the interview by mail. Data Collection Procedure All American Indians with a known address comprised the initial participant pool. Potential participants were called and asked to take part in the current study. All of those who agreed to participate were first sent copies of the interview schedule (with interviewer prompts removed) to allow them to review the content. An interviewer then obtained participant responses by telephone. Individuals who did not have telephones were asked to return the materials by mail. Participants were given $50 to complete the interview. Instruments Life Situation Questionnaire—Revised Version (LSQ-R). The LSQ-R was used to measure multiple aspects of SWB (Krause, 1998a , 1998b ). Eight SWB scales are included in the LSQ-R, seven of which are based on two major sets of subjective items. These two item sets include 20 satisfaction items and 30 problems items. The first item set requires an individual to rate his or her satisfaction with a particular area of life on a 5-point scale. The second set of items requires individuals to rate the degree to which different areas of their lives have caused them problems (1 = no problem, 5 = major problem ). In addition, 2 items ask individuals to rate their overall adjustment to SCI on a 10-point scale, once rating their current adjustment and a second time predicting their adjustment in 5 years. Seven homogeneous scales were developed from factor analysis of the satisfaction and problems items ( Krause, 1998a ): (a) Engagement, (b) Negative Affect, (c) Health Problems, (d) Finances, (e) Career Opportunities, (f) Living Circumstances, and (g) Interpersonal Relations. Higher scores indicate more favorable SWB for five of the seven scales (all except Negative Affect and Health Problems, on which lower scores are more favorable). In addition, summation of the two 10-point items produced an adjustment scale, because previous research suggested that these items tapped a single dimension ( Krause, 1992 ). All scales have been converted to T scores with a mean of 50 and a standard deviation of 10 to allow direct comparisons across scales and to compare samples with those obtained in the development of the LSQ-R. Alpha coefficients for the factor scales ranged from .79 to .92, with an average of .86. Coefficients for the current sample ranged from .73 to .92, with an average of .82. Older Adult Health and Mood Questionnaire. The OAHMQ ( Kemp & Adams, 1995 ) is a 22-item measure of depressive symptoms developed according to the Diagnostic and Statistical Manual of Mental Disorders (revised 3rd edition; American Psychiatric Association, 1987 ). All items involve a true—false format. The instrument was designed to include few physiological ("vegetative") symptoms of depression, because these types of symptoms may parallel the actual sensory and motor complications associated with SCI ( Kendall, Edinger, & Eberly, 1978 ; Taylor, 1967 ). The OAHMQ has been used in investigations of adults of various ages who have a disability and has been found to be valid across age groupings ( Kemp, Adams, & Campbell, 1997 ). The cutoff scores are 0—5 (nondepressed), 6—10 (clinically significant symptomatology), and 11—22 (possible major depression). The OAHMI was validated against clinical assessments by psychiatrists and psychologists and was found to have highly acceptable internal (.87) and test—retest (.84) reliability. The alpha coefficient for the current sample was .91. Reciprocal Social Support Scale. The Reciprocal Social Support Scale ( Anson, Stanwyck, & Krause, 1993 ) was developed to measure amount of support given and received. The scale was modified for the current study, in that participants were asked only to indicate the amount of support received (as opposed to both given and received). Individuals rated the frequency with which they received four types of support: social interaction, material assistance, emotional support, and nonpaid personal assistance. A 7-point scale was used (1 = never, 7 = always ). Participants rated each type of support received from their families, friends, and community. They were also asked the frequency with which upsetting things happened between them and members of their family, their friends, or their community. Five scales were developed, one for each type of support (social interaction, material assistance, emotional support, and nonpaid personal assistance) and one for total upsets; each scale had a score range of 3—21. Alpha coefficients for the current sample ranged from .70 to .76 for the four types of support, with an average of .73. The alpha for the upsets scale was only .55; however, low internal consistency was expected given that the scale sums interactions with three groups of people (family, friends, and community). Supplemental items. Three supplemental sets of items were used. The first item set was identified from the Behavioral Risk Factor Surveillance System (BRFSS; Powell-Griner, Anderson, & Murphy, 1997 ), a survey developed by the Centers for Disease Control and Prevention as a means of gathering information on the health and risk behaviors of adults in each state. The survey is made up of three parts: (a) a standardized core section consisting of questions on health care access, exercise, tobacco use, and other respondent characteristics; (b) optional modules that can be asked, at the state's discretion, on topics such as sexual behavior and injury control; and (c) a section added by an individual state that focuses on a particular topic of interest to that state. Items selected from the BRFSS for this study included self-rated health, number of days per month over the past year in which alcohol was consumed, number of days of poor health in the past month, and number of days in which poor health stopped normal activities. In addition to the BRFSS items, four activity items from the Craig Handicap Assessment and Reporting Technique (CHART; Whiteneck, Charlifue, Gerhart, Overholser, & Richardson, 1992 ) were also used: number of hours per day spent out of bed, average number of days per week spent out of the home, number of nights in the past year spent away from home, and current employment status. Finally, two additional items related to secondary conditions were included: number of secondary injuries within the previous year and number of pressure ulcers within the previous year. Items from the BRFSS and CHART were selected for their potential importance in relation to the SWB outcomes being studied, as well as their widespread use with multiple populations (the BRFSS with nondisabled persons and the CHART with individuals with SCI and other disabilities). The two secondary conditions items were selected due to their potential importance in relation to SWB and their straightforward reporting (i.e., not attempting to measure an underlying construct). Analyses Both descriptive and predictive analyses were implemented for each of the study outcomes. Descriptive procedures used for depression were as follows: Average OAHMQ scores across the full sample were calculated, and individuals were classified into one of three groups based on clinically relevant cutoffs ( Kemp & Adams, 1995 ). Descriptive data were also generated for the eight SWB scales and compared with scale norms via a one-sample t test. Differences between the sample and normative means were converted into standard deviation units (using the difference between the two means, divided by the sample standard deviation). Given the exploratory nature of the study, no Bonferroni correction was used for these eight tests. Although this increased the probability of Type I error, using the correction would have increased the probability of Type II error and potentially masked important relationships as a result of limited power. A two-stage hierarchical (stepwise) multiple regression procedure was used to identify the optimal predictors of OAHMQ and SWB scores (separate analyses were conducted for each scale). There were four biographic and injury-related variables: gender, age at injury onset, years since injury, and cervical versus noncervical injury. Completeness of injury was not included because previous research ( Dijkers, 1997 ) has revealed measures of disability to be minimally related to SWB at best (i.e., injury level was a sufficient indicator). There were 17 explanatory variables: the 4 activity variables from the CHART (work status, hours out of bed per day, average days out of the home per week, and nights spent away from home in the previous year), income and years of education from the CHART, the 5 social support scales (the upsets scale and the 4 types of support scales), 2 secondary conditions variables (injuries and pressure ulcers in the previous year), and the 4 BRFSS items (ratings of health, number of poor health days in the previous month, days adversely affected by poor health in the previous month, and average days per month consuming alcoholic beverages). The 4 biographic and injury-related variables were evaluated during the first stage; a p value of .05 was required for a variable to enter. All other variables were evaluated during the second stage. When the stepwise procedure was used, the predictor that accounted for the greatest portion of variance in study outcomes was selected first, followed sequentially by the variables with the greatest explanatory value (i.e., the portion of the variance not accounted for by the predictors already in the equation). The procedure was implemented until none of the remaining predictors were statistically significant. RESULTS Participant Characteristics Overall, 78% of the participants were male, and 49% reported cervical injuries. In terms of neurologic completeness, 29% reported no sensation or movement below the lesion, 39% reported some sensation with no movement, and 32% reported at least some movement below the level of the lesion (only 13% overall reported functional movement). The average age of the participants was 28.1 years at the time of injury. An average of 12.4 years had passed since the injury (the mean age at the time of the study was 40.4 years). The majority of participants (83.5%) were injured as the result of motor vehicle crashes (only 5% were injured as the result of violence). Forty-nine percent of the participants reported consuming alcoholic beverages within 6 hr preceding the injury (among those who could recall well enough to answer definitively). The participants averaged 11.9 years of education, and 21% were employed at the time of the study. In terms of annual combined family income, 69.1% reported less than $10,000, 19.1% reported between $10,000 and $25,000, 7.4% reported between $25,000 and $40,000, and the remaining 4.4% reported between $40,000 and $55,000. Descriptive Comparisons With SCI Norms The average OAHMQ score was 6.3 ( SD = 5.7). In classifications of participants into three groups based on diagnostic categories, 53% were nondepressed (score of 0—5), 27% reported significant depressive symptomatology (score of 6—10), and the remaining 20% reported probable major depression (score of 11—22). Table 1 summarizes the scores on the eight SWB scales from the LSQ-R, including means, standard deviations, and one-tailed t tests with the LSQ-R norms ( M = 50, SD = 10). Table 2 summarizes data on other variables used as predictors during regression analyses. Significant differences were identified between the American Indian sample and the normative group on five of the eight scales. The most prominent differences were noted on the Career Opportunities scale, on which the American Indians averaged 44.2 (0.58 of a standard deviation below the mean of the normative SCI sample), and the Adjustment scale, on which they averaged 45.3 (0.47 of a standard deviation below the mean). The magnitude of the differences on the remaining three scales in which significant differences were noted was about 2 points, the American Indian sample scoring higher on the Negative Affect (52.5) and Health Problems (52.0) scales and lower scoring on the Finances scale (48.0; see Table 2 for means and standard deviations of nonoutcome variables). Prediction Table 3 summarizes the results of the regression analysis across the nine outcome scales. The multiple correlations ranged from .32 to .77 between the predictors and the scales. The best prediction was for the OAHMI; the predictors accounted for 59% of the variance in scores. The squared multiple correlations for the remaining scales ranged between .10 and .52 ( M = .30). The number of predictors that entered the equation ranged from one (Interpersonal Relations) to eight (Negative Affect). The biographic and injury-related variables failed to enter as predictors in any of the regression analyses. Table 4 provides a complete summary of the regression analyses. Depression. Six explanatory variables entered into the prediction of depression and accounted for 59% of the variance in OAHMQ scores. The first variable to enter the equation was the number of days in which normal activities were stopped during the previous month as a result of health problems (accounting for 39% of the variance). OAHMQ scores were associated with having a greater number of normal activities stopped during the previous month as a result of poor health, consuming alcohol more days per month, leaving the house on fewer occasions, having more upsetting things happen, receiving less personal assistance, and receiving greater material assistance. SWB. The Engagement scale was predicted by four variables (accounting for 34% of the variance). The variable of primary importance was social interaction, which accounted for 21% of the variance in responding. Higher engagement scores were associated with greater support for social interaction, completion of fewer years of education, fewer days disrupted during the previous month as a result of poor health, and fewer upsetting things happening. Eight scales entered into the prediction of negative affect and accounted for 52% of the variance. Higher negative affect was associated with a greater number of days of normal activities stopped as a result of poor health, more total upsets, having less personal assistance support and less income, having completed more years of education, having more average days drinking per month, receiving more material aid, and leaving the home on a fewer number of days per week. Only two variables entered into the prediction of health problems (accounting for 30% of the variance). Health problems were associated with lower self-rated general health and a greater number of days in which poor health prevented normal activity in the previous month. In contrast, four variables significantly entered into the prediction of finances but accounted for only 23% of the variance in Finances scale scores. Higher scores were associated with fewer days in which normal activities were stopped as a result of poor health, fewer upsetting events, more personal support, and more material aid. Five variables significantly added to the prediction of career opportunities, accounting for 39% of the variance. Career Opportunities scale scores were associated with fewer days of normal activity stopped within the previous month as a result of poor health, current employment, fewer days per month drinking alcohol, more support for social interaction, and more nights away from home per year. Three variables added to the prediction of living circumstances, accounting for 31% of the variance in responses. Living Circumstances scale scores were associated with greater emotional support but fewer total upsets and fewer days of drinking per month. Only one variable, emotional support, aided the prediction of interpersonal relations (accounting for 10% of the variance in responses). Greater emotional support was related to higher Interpersonal Relations scale scores. On the final scale, Adjustment, four variables significantly added to the predictive equation (accounting for 34% of the variance in adjustment scores). Higher adjustment scores were associated with overall higher ratings of health, more days spent per week out of the house, fewer days spent drinking per month, and more emotional support. DISCUSSION The current study represented one of the first attempts, if not the first attempt, to identify psychological outcomes among American Indians with SCI. Both clinically relevant outcomes (i.e., depression) and nonclinical outcomes (i.e., SWB) were examined. Specifically, this study was intended to identify the types of injury-related, health-related, activity-related, and social support factors that are associated with the degree to which individuals attain reasonable adjustment or SWB after SCI. Comparisons With Other SCI Populations Although it is not always valid to compare scores between studies, the results of the current study with American Indians suggest that this population may be at greater risk for adverse clinical and nonclinical psychological outcomes after SCI. Specifically, the percentage of our participants who reported clinically significant symptomatology for depression (47%) exceeded those observed for Caucasians (30%) and African Americans (36%) in a separate study but roughly equaled that of Latinos (46%; Kemp et al., 1999 ). Several aspects of SWB also involved scores that were significantly different from those of the normative population. Career Opportunities scale scores were more than half a standard deviation below the normative means, with similar differences in Adjustment scale scores relative to previous research ( Krause, 1998a ). Other scores were also significantly different from that of the instrument norm, including Negative Affect, Health Problems, and Finances scores (ranging from 0.2 to 0.25 standard deviations). These results reflect meaningful (although generally not large) differences between American Indians with SCI and the normative group composed of Caucasians and African Americans. African Americans also reported lower scores for finances and career opportunities. However, this American Indian sample appeared to report as well poor general outcomes related to negative affect and health problems (American Indians were not different from Caucasians on these latter two aspects of SWB). Regression Results One of the more interesting findings of the current study was that biographic and injury-related variables were not associated with the SWB outcomes that were the focus of this study (either clinically or nonclinically related). Although it is not surprising that these variables were not of great magnitude ( Dijkers, 1997 ), it is important to note that they did not enter into the equation for a single outcome, despite being forced into the equation before the other sets of variables. Therefore, at least among American Indians with SCI, biographic and injury-related variables do not appear to be important correlates of psychological outcomes. Because biographic and injury-related factors were not important in regard to psychosocial outcomes, it becomes important to look at other life areas to understand factors that are associated with more or less favorable SWB after SCI among American Indians. The types of factors that were investigated in this study, outside of the basic biographic and injury-related variables, fall into five categories: (a) general activities (from the CHART), (b) education and income, (c) social support, (d) general health (from the BRFSS), and (e) the two types of secondary conditions (injuries and pressure ulcers). It appears that health-related variables were the most important correlates of a diverse number of psychological outcomes among the study population, including both clinically relevant outcomes (depression and negative affect) and other aspects of SWB. In fact, health factors from the BRFSS were the first variables to enter the equation for six of the nine outcomes (all but engagement, living circumstances, and interpersonal relations). The type of specific health variable most strongly associated with SWB varied as a function of study outcome. However, days adversely affected by poor health was the variable most strongly associated with SWB overall and the first variable to enter for negative affect, finances, and career opportunities. Overall ratings of health were primarily important in regard to the Health Problems and Adjustment scales, whereas poor health days were primarily important in terms of the OAHMQ and Health Problems scales. It is noteworthy that neither of the two secondary conditions (injuries or pressure ulcers) entered into the prediction of any of the psychological outcomes. In sum, it appears that general health, particularly the extent to which poor health does not interfere with day-to-day activities, is the primary factor associated with favorable psychological adjustment among American Indians with SCI. Social support was also highly predictive of psychological outcomes, although there appears to be a closer match between the type of support provided and the type of psychological outcome. For example, emotional support was strongly associated with two SWB variables, both of which were consistent with the need to preserve a warm and emotionally supportive environment (living circumstances and interpersonal relations). In contrast, support related to social interaction was associated with only two variables, engagement and career opportunities, both of which intuitively would be related to social interaction. It is noteworthy that amount of nonpaid personal assistance was favorably related to multiple psychological outcomes, whereas material assistance was generally negatively related to sound psychological functioning (particularly clinically relevant outcomes). Number of upsetting events was related to the most diverse set of outcomes of any of the social support factors (depression, engagement, finances, and living circumstances). Although income was negatively correlated with negative affect, as might be expected, it was surprising that years of education was positively correlated with negative affect and negatively correlated with engagement. It was also surprising that neither income nor years of education entered into the prediction of the SWB scales related to finances or career opportunities (as hypothesized). Activities were not highly correlated with the psychological outcomes that were the focus of the current investigation. In fact, hours spent out of bed did not contribute to prediction of any study outcomes. Work status entered only into the prediction of career opportunities, as would be expected. Getting out of the home appeared to be highly related to avoiding depression and negative affect, as well as being positively associated with more favorable self-ratings of adjustment. Finally, use of alcohol deserves special attention. Number of days using alcohol entered the equation for five separate outcomes. These diverse outcomes were depression, negative affect, career opportunities, living circumstances, and self-rated adjustment. Therefore, at least among this population, use of alcohol was associated with poorer clinical and nonclinical outcomes. Implications There are several implications of this study for the rehabilitation of American Indians with SCI. First, depression and diminished SWB do appear to be issues that may be particularly important among this population. Therefore, it is necessary to develop means of intervening in cases of depression, as well as prevention strategies that minimize the development of emotionally related problems. Second, lower SWB in certain areas, particularly career and finances, may relate to restricted opportunities among American Indians. The unexpected positive association between years of education and negative affect supports this assertion, along with the finding that neither income nor education predicted SWB related to finances or career opportunities. In short, education may not lead to anticipated benefits of increased income and quality of life for American Indians with SCI. This may reflect the need for societal or legislative protection for individuals in this racial—ethnic minority group, as well as other similar groups that have been found to report similar SWB in these areas (e.g., African Americans). Special efforts may be needed to equalize opportunities for such individuals. Third, because social support was highly related to many psychological outcomes, it may be particularly important to build support networks as early as possible after SCI. Many American Indians return to isolated environments, such as reservations; thus, it may be important to work with these environments to attempt to bolster support as much as possible while individuals are still receiving initial rehabilitation in the hospital environment. Given the decreasing lengths of stay after SCI, these support networks ideally must be built in the areas where the individuals reside. Finally, alcohol issues are clearly associated with poor outcomes among this population (this finding may generalize to others as well). Therefore, as is the case among other groups with SCI, any attempt at intervention during the initial rehabilitation period may be important to facilitate long-term outcomes and to avoid unnecessary development of depression or low SWB. Limitations Although one of the primary strengths of the study was the focus on American Indians, this limits generalizability to other populations. In other words, whereas previous research has limited generalizability in regard to American Indians, the current study involves limited generalizability, or at least unknown generalizability, to more general populations. Even among American Indians, the current study generalizes mostly to those in the western United States (particularly Navajo, Zuni, and Sioux). Furthermore, all data were based on self-report. Objective measures of health may have produced different findings, although this is far from certain. Another limitation is that this study used variables that were reflective of health, secondary conditions, activities, and social support. Therefore, it did not use personality measures in conjunction with psychological outcomes or other types of psychometric instruments that may have enhanced the overall prediction. Moreover, this study was basically atheoretical and did not investigate cultural factors per se. Although this was a result of the exploratory nature of the present research (basically the first such study of American Indians with SCI), future investigations will benefit from incorporating theoretical and cultural variables into the research design. Finally, because all data were cross sectional, there is no way to determine the direction of causation (if any). Future Research Clearly, there is a need to increase the amount of research on American Indians with SCI and other disabilities. Although the current findings do not definitively indicate that American Indians have a different profile for the types of factors that predict psychological outcomes, it does appear that their outcomes are generally worse than those associated with Caucasians. It is also important to implement these studies to ensure generalizability to the American Indian population with SCI. More research is needed as well to identify the specific types of cultural factors underlying the relationships identified in this study. It may be particularly important to identify the roles of education, poverty, and residence (reservation or nonreservation) in quality of life. Furthermore, treatment and prevention studies of poor psychological outcomes after SCI, both among American Indians and among the general SCI population, need to be implemented. Finally, there is a need to investigate health outcomes in a manner similar to the psychological outcomes in the current study. It will take multiple efforts, both research based and clinically based, for American Indians to have equal opportunities to obtain successful psychological and health outcomes after SCI, as well as full participation in their chosen activities. References American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders ( (rev. 3rd ed.). Washington, DC: Author) Anson, C. A., Stanwyck, D. J. & Krause, J. S. (1993). Social support and health status in spinal cord injury. International Journal of Paraplegia, 31, 632-638. Carlson, C. E. (1979). Conceptual style and life satisfaction following spinal cord injury. Archives of Physical Medicine and Rehabilitation, 60, 346-352. Cheadle, A., Pearson, D., Wagner, E., Psaty, B. M., Diehr, P. & Koepsell, T. (1994). Relationship between socioeconomic status, health status, and lifestyle practices of American Indians: Evidence from a Plains reservation population. Public Health Reports, 109, 405-413. DeVivo, M. J. & Fine, P. R. (1982). Employment status of spinal cord injured patients 3 years after injury. Archives of Physical Medicine and Rehabilitation, 63, 200-203. DeVivo, M. J. & Richards, J. S. (1992). Community reintegration and quality of life after spinal cord injury. Paraplegia, 30, 108-112. DeVivo, M. J., Rutt, R. D., Stover, S. L. & Fine, P. R. (1987). Employment and spinal cord injury. Archives of Physical Medicine and Rehabilitation, 68, 494-498. Dijkers, M. (1997). Quality of life after spinal cord injury: A meta-analysis of the effects of disablement components. Spinal Cord, 35, 829-840. Fuhrer, M. J., Rintala, D. H., Hart, K. A., Clearman, R. & Young, M. E. (1993). Depressive symptomatology in persons with spinal cord injury who reside in the community. Archives of Physical Medicine and Rehabilitation, 74, 255-260. Goldberg, H. I., Warren, C. W., Oge, L. L., Friedman, J. S., Helgerson, S. D., Pepion, D. D. & LaMere, E. (1991). Prevalence of behavioral risk factors in two American Indian populations in Montana. American Journal of Preventive Medicine, 7, 155160. James, M., DeVivo, M. J. & Richards, J. S. (1993). Postinjury employment outcomes among African-American and White persons with spinal cord injury. Rehabilitation Psychology, 38, 151-164. Kemp, B. J. & Adams, B. M. (1995). The Older Adult Health and Mood questionnaire: A measure of geriatric depressive disorder. Journal of Geriatric Psychiatry and Neurology, 8, 162-167. Kemp, B. J., Adams, B. M. & Campbell, M. L. (1997). Depression and life satisfaction in aging polio survivors versus age-matched controls: Relation to postpolio syndrome, family functioning, and attitude toward disability. Archives of Physical Medicine and Rehabilitation, 78, 187-192. Kemp, B. J. & Krause, J. S. (1999). Depression and life satisfaction among people aging with a disability: A comparison of post-polio and spinal cord injured individuals. Disability and Rehabilitation, 21, 241-249. Kemp, B. J., Krause, J. S. & Adkins, R. A. (1999). Depressive symptomatology among African-American, Latino, and Caucasian participants. Rehabilitation Psychology, 44, 235-247. Kendall, P., Edinger, J. & Eberly, C. (1978). Taylor's MMPI correction factor for spinal cord injury: Empirical endorsement. Journal of Consulting and Clinical Psychology, 46, 370-371. Krause, J. S. (1992). Longitudinal changes in adjustment after spinal cord injury: A fifteen-year study. Archives of Physical Medicine and Rehabilitation, 73, 564-569. Krause, J. S. (1997). Adjustment after spinal cord injury: A nine-year longitudinal study (1985—1994). Archives of Physical Medicine and Rehabilitation, 78, 651-657. Krause, J. S. (1998a). Changes in adjustment after spinal cord injury: A 20-year longitudinal study. Rehabilitation Psychology, 43, 41-55. Krause, J. S. (1998b). Dimensions of subjective well being after spinal cord injury: An empirical analysis by gender and race/ethnicity. Archives of Physical Medicine and Rehabilitation, 79, 900-909. Krause, J. S. (1998c). Subjective well being after spinal cord injury: Relationship to gender, race/ethnicity, and chronologic age. Rehabilitation Psychology, 43, 282-296. Krause, J. S. & Anson, C. A. (1997a). Adjustment after spinal cord injury: Relationship to gender and race. Rehabilitation Psychology, 42, 31-46. Krause, J. S. & Anson, C. A. (1997b). Adjustment after spinal cord injury: Relationship to participation in employment or educational activities. Rehabilitation Counseling Bulletin, 40, 202-214. Krause, J. S., Coker, J. L., Charlifue, S. W. & Whiteneck, G. G.Health behaviors among American Indians with spinal cord injury: Comparison to data from the 1996 Behavioral Risk Factor Surveillance System. Archives of Physical Medicine and Rehabilitation, , (in press) Krause, J. S. & Crewe, N. M. (1991). Chronologic age, time since injury, and time of measurement: Effect on adjustment after spinal cord injury. Archives of Physical Medicine and Rehabilitation, 72, 91-100. Krause, J. S., Sternberg, M., Maides, J. & Lottes, S. (1998). Employment after spinal cord injury: Differences related to geographic region, gender, and race. Archives of Physical Medicine and Rehabilitation, 79, 615-624. Lowe, L. P., Long, C. R., Wallace, R. B. & Welty, T. K. (1997). Epidemiology of alcohol use in a group of older American Indians. Annals of Epidemiology, 7, 241248. Miller, N. B. (1978). Utilization of services for the developmentally disabled by American Indian families in Los Angeles. Dissertation Abstracts International, 39, 3016A Morgan, J. & O'Connell, J. C. (1987). The rehabilitation of disabled Native Americans. International Journal of Rehabilitation Research, 10, 139-149. Pearson, D., Cheadle, A., Wagner, E., Tonsberg, R. & Psaty, B. M. (1994). Differences in sociodemographic, health status, and lifestyle characteristics among American Indians by telephone coverage. Preventive Medicine, 23, 461-464. Peterson, D. E., Remington, P. L., Kuykendall, M. A., Kanarek, M. S., Diedrich, J. M. & Anderson, H. A. (1994). Behavioral risk factors of Chippewa Indians living on Wisconsin reservations. Public Health Reports, 109, 820-823. Powell-Griner, E., Anderson, J. E. & Murphy, W. (1997). State- and sex-specific prevalence of selected characteristics: Behavioral Risk Factor Surveillance System, 1994 and 1995. Morbidity and Mortality Weekly Reports, 46(3), 1-31. Sugarman, J. R., Warren, C. W., Oge, L. & Helgerson, S. D. (1992). Using the Behavioral Risk Factor Surveillance System to monitor Year 2000 objectives among American Indians. Public Health Reports, 107, 449-456. Taylor, G. (1967). Predicted versus actual response to spinal cord injury: A psychological study. ( Unpublished doctoral dissertation, University of Minnesota, Minneapolis) Welty, T. K., Lee, E. T., Yeh, E., Cowan, L. D., Go, O., Fabsitz, R. R., Le, N. A., Oopik, A. J., Robbind, D. C. & Howard, B. V. (1995). Cardiovascular disease risk factors among American Indians: The Strong Heart Study. American Journal of Epidemiology, 142, 269-287. Whiteneck, G. G., Charlifue, S. W., Gerhart, K. A., Overholser, J. D. & Richardson, G. N. (1992). Quantifying handicap: A new measure of long-term rehabilitation outcomes. Archives of Physical Medicine and Rehabilitation, 73, 519-526. Wilson, C., Civic, D. & Glass, D. (1995). Prevalence and correlates of depressive syndromes among adults visiting an Indian Health Services primary care clinic. American Indian and Alaskan Native Mental Health Research, 6(2), 1-12. This research was supported by Model Spinal Cord Injury Systems grants from the National Institute for Disability and Rehabilitation Research of the Office of Special Education and Rehabilitative Services, U.S. Department of Education (H133N00023 and H133N50001). We would like to thank the following people, without whose contributions completion of this article would not have been possible: Linda Singer, from Craig Hospital, who conducted all the interviews, and Else Henry, Samuel Jones, Sarah Lottes, and Kyle Massey from the Shepherd Center. Correspondence may be addressed to J. Stuart Krause, Shepherd Center, Crawford Research Institute, 2020 Peachtree Road NW, Atlanta, Georgia, 30309. Received: December 4, 1998 Revised: July 6, 1999 Accepted: July 7, 1999 Table 1. Comparison of Sample Means of American Indians With Life Situation Questionnaire—Revised Version Normative Scores Table 2. Means and Standard Deviations of Independent Variables Table 3. Summary of Regression Analyses: Number of Predictors, Multiple and Squared Multiple Correlations, and F Values Table 4. Regression Summary for Each Scale